Finomnis

I would request changes before merge. The direction is useful, and the sample/PID cleanup looks mostly aligned, but the current implementation has several correctness and robustness problems in host-controlled parsing, shared state, identity handling, buffer sizing, and test coverage.

This is a static review of the PR diff and discussion; I did not run this on USB hardware.

What looks good

The PR now uses the next Zephyr sample PID, adding usb-mtp as 0x0012 in the USB PID table and setting the sample PID to 0x0012. That resolves the earlier PID collision directionally. ([GitHub][1])

The public API is small and understandable: USBD_MTP_DEFINE_STORAGE(id, mountpoint, read_only) registers filesystem-backed MTP storage via iterable sections, and the sample demonstrates a RAM-backed FATFS storage mounted at /RAM:. ([GitHub][1])

Finomnis

public class Test {
	
	private static class A{
		public static void test(){
			System.out.println("A:test");
		}
		
		void dynamic(){
			System.out.println("A:dynamic");
		}

Finomnis

===== 3-gon =====
    10 (3, 4)
    120 (8, 15)
    1540 (20, 55)
    7140 (34, 119)

===== 4-gon =====
    4900 (24, 70)

Finomnis

#!/usr/bin/env python3
import math
import time


def isPrime(n):
    max_divisor = math.floor(math.sqrt(n))
    for d in range(2, max_divisor+1):
        if n % d == 0:
            return False

Finomnis

This is how our current reduce algorithm works.

A A A A A A A A

split:
A A A A   A A A A

accumulate:
           A  A  A  A
initial B->+->+->+->+->B